As transistor geometry continues to shrink, more and more functional circuitry may be embedded within integrated circuits (ICs). This trend is beneficial for the electronics industry since it enables development of smaller, lower power electronic consumer products, such as cell phones and hand held computers. However, as IC circuit density increases, the testing of ICs becomes more complex and costly for the IC manufacturers. Reducing the cost of manufacturing ICs is a primary goal for every IC manufacturer. By reducing IC manufacturing cost, an IC manufacturer can advantageously cost-differentiate its IC products from other IC manufacturers.
FIG. 1A illustrates a semiconductor wafer 101 comprising multiple die 102 circuits. FIG. 1B illustrates one of the die circuits 101 on wafer 101. The die contains core circuitry 103, which provides the functionality of the die, and pad locations 104 for providing contacts for accessing the core circuitry.
FIG. 1C illustrates a conventional test arrangement for contacting and testing a single die 102 of wafer 101. The test arrangement includes a tester 105, a single die probe mechanism 109, and a die 102 to be tested. Tester 105 comprises a controller 106, stimulus circuitry 108, and response circuitry 107. Controller 106 regulates the stimulus circuitry 108 via interface 117 to output test stimulus signals to die 102 via stimulus bus 111. Controller 106 regulates the response circuitry 107 via interface 118 to receive test response signals from die 102 via response bus 110. Probe mechanism 109 comprises the stimulus bus 111 and response bus 110 connection channels between tester 105 and die 102. The probe mechanism contacts the input 115 and output 116 die pads via small probe needles 112. While only a pair of input and output probe needles 112 are shown in this simple illustration, it is understood that all die input and output pads will be similarly contacted by the probe mechanism 109 using additional probe needles 112. The input pads 115 transfer stimulus signals to core 103 via input buffers 113, and the output pads 116 transfer test response signals from core 103 via output buffers 114. The testing of the die 102 in FIG. 1C occurs through the process of inputting stimulus signals to the die and receiving response signals from the die.
FIG. 2 illustrates in more detail the stimulus 108 and response 107 circuitry of tester 105. Stimulus circuitry 108 typically comprises a large stimulus data memory 201 for storing the stimulus data to be applied to the die. Controller 106 controls the loading of the stimulus data memory 201 from a source, such as a hard disk, prior to testing, and then controls the stimulus data memory to output the loaded stimulus data to the die during test, via stimulus bus 111. Response circuitry 107 typically comprises a large mask and expected data memory 203, a comparator 204, and a fail flag memory 202. The mask and expected data memory 203 stores mask and expected data to be used by the comparator 204 to determine if the response data from the die passes or fails.
During test, the comparator 204 inputs response signals from the die via response bus 110, and mask (M) and expected (E) data signals from memory 203 via mask and expected data buses 206 and 207. If not masked, by mask signal input from memory 203, a given response signal from the die is compared against a corresponding expected data signal from memory 203. If masked, by mask signal input from memory 203, a given response signal from the die is not compared against an expected data signal from memory 203. If a non-masked response signal matches the expected signal, the compare test passes for that signal. However, if a non-masked response signal does not match the expected signal, the compare test fails for that signal and the comparator outputs a corresponding fail signal on bus 205 to fail flag memory 202. At the end of test, the controller 106 reads the fail flag memory to determine if the die test passed or failed. Alternately, and preferably in a production test mode, the single die test may be halted immediately upon the controller receiving a compare fail indication from the fail flag memory 202, via the interface 118 between controller 106 and response circuitry 107, to reduce wafer test time. At the end of the single die test, the probe mechanism is relocated to make contact to another single die 102 of wafer 101 and the single die test is repeated. The wafer test completes after all die 102 of wafer 101 have each been contacted and tested as described above.
FIG. 3 illustrates a conventional test arrangement for simultaneously contacting and testing multiple die 102 of wafer 101. The test arrangement includes tester 105, multiple die probe mechanism 301, and a multiple die 1-N 102 to be tested. The difference between the single and multiple die test arrangements of FIGS. 2 and 3 is in the use of the multiple die probe mechanism 301. As seen in FIG. 3, the connection between probe mechanism 301 and tester 105 is as previously described. However, the connection between probe mechanism 301 and die 1-N is different. Each stimulus bus signal from the tester uniquely probes common pad inputs on each die 1-N. For example, the stimulus 1 (S1) signal from the stimulus bus probes all common input pads 303 of all die 1-N via connection 302. While not shown, stimulus 2-N (S2-N) signals from the stimulus bus would each similarly probe all other common input pads of all die 1-N. This allows the stimulus bus signals to simultaneously input the same stimulus to all die 1-N during the test.
As seen in FIG. 3, the die response connection of probe mechanism 301 is different from the above described die stimulus connection. Whereas each common input pad 303 of die 1-N share a single stimulus signal connection 302, each common output pad 304 requires use of a dedicated response signal connection. For example, output pad 304 of die 1 uses a response signal connection 305, output pad 304 of die 2 uses a response signal connection 306, output pad 304 of die 3 uses a response signal connection 307, and output pad 304 of die N uses a response signal connection 308. All other output pads of die 1-N would similarly use a dedicated response signal connection. All dedicated response signal connections are channeled into the response bus to tester 105, as seen in FIG. 3. During test, the tester outputs stimulus to all die 1-N and receives response outputs from all die 1-N. The test time of testing multiple die in FIG. 3 is the same as testing single die in FIG. 2. The test operates in the masked/non-masked compare mode as described in FIGS. 1C and 2. When testing multiple die simultaneously, as opposed to testing a single die, a production test preferably runs to completion even though an early compare may occur on one or more of the die being tested. This is done because typically most of the die will pass the production test and aborting the multiple die production test on a failure indication would actually increase the test time, since the test would need to be re-run later to complete the testing of the passing die.
The limitation of the multiple die test arrangement in FIG. 3 lies in the number of dedicated response inputs 305-308 the tester 105 can accept on its response bus. For example, if the tester can accept 300 response input signals and each die has 100 output pads, the multiple die test arrangement of FIG. 3 is limited to only being able to test 3 die at a time. Testing 300 die on a wafer with this 3 die per test limitation would require having to relocate the probe mechanism 301 approximately 100 times to contact and test three die at a time. The time required to relocate the probe mechanism and repeat the die test say 100 times consumes test time which increases the cost to manufacture the die. It is possible to widen the response bus input of the tester to say 600 inputs to allow testing 6 die at a time, but adding circuitry to the tester to increase its response bus input width is expensive and that expense would increase the cost of manufacturing die.